I have a very large stress analysis model which force me to split the same into two separate models due to the time it takes to run as a single one.
My question is in this case how to define the boundary conditions in split models. I have separated the models at the location where i had a limit stop (+x,-x,+y,+z & -z).

I have done an exercise in which i separated the models at the location of limit stop and analyzed both the systems separately.global force and moments at the limit stop location in each model is found and the same is kept at the limit stop node of other model. However the loads on the restraints in the split models is not the same as the one i got in the initial single model.

What can be the solution for this problem ?

Of course the best place to split the models is at an anchor. If you wish to split the model at any other location it will be up to you to establish equilibrium at your break point. Your +x,-x, +y, +z & -z is a good start but you must watch your rotation terms. All 6 degrees of freedom at this termination should provide equal and opposite loads to keep the break point in equilibrium.
Are you sure there are no interior anchors to use?

Agreed with Dave..
ANCHOR is the best location to separate any large stress system. Rotation will create different set of loads on the nearby supports.

Moreover, If your system doesn't have any anchor then at least model the system upto first guide from the 3D stop.

Dave, please correct me if I am wrong.